qm_mm/image_charge: optical spectrum

[Victor Volkov]

Good day.

I wish to bring under attention 

of the developers 

(of this approach

https://manual.cp2k.org/trunk/methods/qm_mm/image_charges.html)

two examples, 

where 

their code, apparently, fails to compute

optical electronic transitions.

Specifically, following the online examples,

I condition OH neutral molecule at four gold layers (facet 111),

where gold atoms are fixed under hexagonal symmetry

to provide a stable, well-defined structural reference.

The neutral OH molecule has one unpaired electron.

Therefore multiplicity of the system is 2.

My coworker and me, we run time-dependent DFT to compute three transitions

under regular setting

and under SAOP setting.

Here, I attach inputs and outputs for both cases 

and the coordinate file.

If you wish to try to run the jobs yourself,

you would need Au.pot file, which is present on the CP2K website.

and you would need to comment out the restart wavefunction 

and reset charges to ATOMIC. 

My co-worker and me tested the jobs 

under linux cp2k of different installations,

including the most recent one.

The results are the same:

1) SAOP job dies with the same

 *** WARNING in qs_tddfpt2_utils.F:634 :: Orbital energy correction ***
 *** potential is an experimental feature. Use it with extreme care ***

2) regular TDDFT job dies without any notice.

If possible, 

would you be kind to look at the case and comment,

if the failure(s) can be amended.

If the CP2K QMMM-IMAGE theory cannot solve such cases,

would you instruct kindly

if the recent CP2K version got an update for TD-DFT

to compute optical transitions 

to include only a narrow sub-set of orbitals, which may be selected

to be specific to HOMO-LUMO properties of the adsorbed molecule

rather than to the metal continuum sp-d transitions.   

If such TD-DFT is possible, would you be kind to hint

how to instruct this in input file.

Thank you for your time and attention.

Sincerely, 

I wish you a wonderful and peaceful time 

during the spring festival of life revival.

Victor

[jgh]

Hi

you are testing your luck here. The combination of this many non-standard options is prone to errors.

QM/MM image charges

hexagonal grid

2d PBC

TDDFT

XC grid smoothing

...

When I remove the grid smoothing and run the input with 3d PBC it works (latest trunk version of CP2K), see attached input file.

If the results can be trusted is another question.

Using the SAOP optin also run through but had negative excitation energies.

I wouldn't use SAOP in a production run without first do extensive tests on small systems.

The new TDDFT options to use a restricted occupied orbital set for excitation works for the few systems we have tested up to now.

If you want to try it, please keep in mind that it is a new option.

regards

JH

[Victor Volkov]

Sehr geehrter Dr. Hutter:
Danke vielmal !!

Mit freundlichen Grüßen, 

Victor

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/05bf4e0a-fc52-4223-907e-41b7c3139747n%40googlegroups.com.